Guardian-Edge Microsystems
Problem Solution Technology Roadmap Contact
Ultra-low-power multimodal sensing SoC for safer industrial and energy assets.
GuardianEdge integrates gas, acoustic and vibration sensing, TinyML and event-driven power management into a single SoC that can run for years on a coin cell.
Semiconductors & Advanced Electronics
Multimodal sensing
TinyML at the edge
Hello Tomorrow Global Challenge application – non-confidential overview
Design targets at a glance
Always-on current
< 2–3 µA @ 3 V
Detection latency
< 10–50 ms
Energy per inference
3–5× lower vs. generic MCU (target)
Modalities
Gas + acoustic + vibration / environmental
Architecture validated in MATLAB/Simulink with a multimodal gas + acoustic demo and HDL-ready DFE blocks.

Problem & impact

Refineries, chemical plants and substations still rely on a patchwork of generic MCU nodes, discrete gas detectors and separate vibration or acoustic sensors.

  • Gas detectors are often far from the actual leak – alarms trigger late or not at all.
  • False alarms and nuisance trips lead to alarm fatigue and raised thresholds.
  • Generic MCUs running TinyML in software burn too much energy per inference, limiting sampling rates and analytics richness.

GuardianEdge aims to:

  • Enable earlier detection of leaks and anomalies using fused gas + acoustic + vibration features.
  • Support dense deployment of battery-powered nodes with multi-year coin-cell lifetime.
  • Reduce unplanned downtime and maintenance truck-rolls while improving worker safety.

Solution – GuardianEdge SoC

GuardianEdge is a domain-specific multimodal sensing SoC that integrates:

  • Configurable analog front-ends for gas and environmental sensors.
  • Acoustic/vibration front-ends for machine and infrastructure monitoring.
  • Fixed-point digital front-ends, feature extraction and a TinyML/rule engine.
  • A deeply event-driven power and clock-management state machine.

The SoC is radio-agnostic and connects cleanly to sub-GHz, NB-IoT/LTE-M, Wi-Fi or 5G modems.

Multimodal sensing
Event-driven FSM
TinyML-ready
Radio-agnostic

In our current MATLAB/Simulink proof-of-concept, gas and acoustic signals are processed through dedicated digital front-ends (Gas_DFE and Acoustic_DFE). Each channel computes a severity level; a fusion engine combines them into a fused severity and a critical_alert signal, which drives the future hardware power FSM and alert outputs.

Technology & verification

Core technology

  • MATLAB/Simulink models for gas and acoustic channels, including synthetic leak profiles and machine noise.
  • Fixed-point digital front-ends designed to map cleanly to HDL (Gas_DFE, Acoustic_DFE).
  • Multimodal fusion logic with tunable thresholds and severity levels.
  • Architecture designed for separate always-on and active domains for deep event-driven operation.

Verification & safety roadmap

  • Planned UVM-based, constrained-random and coverage-driven verification at SoC/IP level.
  • Functional & code coverage plus assertion-based verification of key safety properties.
  • Design-for-test: scan insertion, memory and logic BIST, analog loopbacks.
  • Architected to support IEC 61508 / ISO 13849 system-level safety, with EMC and cybersecurity standards (IEC 61010 / 61326 / 62443) and secure boot / hardware root-of-trust on the roadmap.

Technology & product roadmap (0–24 months)

0–6 months – Models & FPGA
  • Finalize MATLAB/Simulink and fixed-point models.
  • Implement HDL-compatible Gas_DFE and Acoustic_DFE.
  • Bring up initial FPGA prototype and demo node.
6–12 months – SoC micro-architecture
  • Refine GuardianEdge SoC micro-architecture.
  • Run synthesis and PnR experiments at 22/28 nm.
  • Integrate TinyML engine on a soft-core MCU and start pilot discussions.
12–18 months – First silicon
  • First MPW tape-out of GuardianEdge GE-SoC v1.
  • Evaluate silicon and system performance.
  • Run field pilots using reference modules.
18–24 months – Deploy & scale
  • Initial commercial deployments.
  • IP licensing discussions with MCU/SoC vendors.
  • Expansion of the TinyML/SDK ecosystem.

Business model & target customers

Platform business model

  • SoC & IP – GuardianEdge SoC sales; licensing key IP blocks such as AFE, DSP and power manager to MCU/SoC vendors.
  • Modules & reference designs – pluggable sensing modules for OEMs and system integrators.
  • Software / SDK / services – configuration tools, TinyML deployment and design-in support for key accounts.

Target customers

  • Industrial gas and safety detection OEMs.
  • Predictive maintenance and condition-monitoring vendors.
  • Energy and infrastructure operators (substations, pipelines, storage).
  • Building automation and industrial system integrators.

Future features – 5G/6G & ISAC

Short term – 5G/IoT-ready endpoint

  • GuardianEdge acts as a radio-agnostic sensing endpoint behind sub-GHz, NB-IoT/LTE-M, Wi-Fi or 5G modems.
  • Nodes transmit compact multimodal feature vectors rather than raw signals.
  • Edge and cloud platforms perform higher-level ISAC analytics and digital twin modelling.

Long term – integrated sensing and communication

  • Add ISAC-DSP blocks to ingest RF features (e.g. channel state information) from 5G/6G radios.
  • Fuse RF with gas/acoustic/vibration data for occupancy mapping and dynamic safety zones.
  • Vision: GuardianEdge becomes the multimodal sensing backbone for 5G/6G-enabled factories and critical infrastructure.

Team & traction

Tai Vo – Founder, SoC and Edge Sensing Lead
M.Sc. Electronics & Nanotechnology, Aalto University, Finland

Senior SoC engineer working on advanced IP and various System-on-Chip projects. Experience from RTL to GDS and architecture of low-power, high-speed designs. Previous recognition in global deep-tech and innovation competitions.

Vi – Co-founder – Bio and Sensing (Germany)
M.Sc. candidate in Biotechnology, Germany

Background in molecular biotechnology, baculovirus expression systems and biosensing. Provides a path to future bio-safety and lab-sensing variants of GuardianEdge.

Proof of concept
MATLAB/Simulink multimodal demo (gas + acoustic) + HDL-ready DFE blocks
Next milestones
FPGA prototype, partner pilots, first MPW tape-out on mature CMOS node

Contact & collaboration

We are looking for industrial safety, energy and infrastructure partners interested in:

For more details or to discuss potential collaboration, please reach out at:
Contact: taipvo993@gmail.com